DP-900 Skills Measured

Structured data has a fixed schema, semi-structured data has flexible fields, and unstructured data does not fit a table cleanly. This distinction drives whether a scenario points toward a relational database, document database, data lake, or file/object store.

Transactional systems process day-to-day business events such as orders, payments, and updates. Analytical systems summarize and explore large datasets for reporting or decision-making. DP-900 often uses OLTP and OLAP clues to separate operational databases from warehouses, lakes, and BI tools.

Know tables, rows, columns, primary keys, foreign keys, joins, normalization, views, indexes, and basic SQL operations. A relational answer usually fits scenarios needing consistent records, relationships between entities, constraints, and familiar query patterns.

Azure SQL Database is the managed relational service most candidates should recognize first. Azure Database for PostgreSQL and Azure Database for MySQL support open-source relational engines. Managed relational services reduce infrastructure management while preserving relational query and schema concepts.

Document, key-value, graph, and column-family models solve different problems. A document model fits flexible JSON records. A graph model fits relationship traversal. A key-value pattern fits simple lookup. Avoid choosing a relational service when the scenario emphasizes flexible schema or globally distributed document data.

Azure Cosmos DB is a globally distributed NoSQL database service. For DP-900, focus on API/model recognition, partitioning at a high level, consistency choices, and why distributed document workloads may not fit traditional relational design.

A data warehouse stores curated, structured data for reporting and analysis. A data lake stores raw or varied files at scale. Lakehouse concepts combine file-based storage with table features that support analytics. The exam often tests the difference between storing raw data and serving modeled reporting data.

Batch processing handles data in scheduled or grouped loads. Streaming handles continuous events such as telemetry, transactions, or device messages. Choose streaming concepts when the scenario emphasizes near-real-time arrival and batch concepts when the data is processed in planned intervals.

Power BI helps create semantic models, reports, dashboards, and visual analysis. Microsoft Fabric appears at a foundational level as a unified analytics platform that includes experiences for data integration, engineering, warehousing, real-time analytics, and BI. DP-900 usually tests recognition, not deep implementation.

Most DP-900 questions are solved by matching the requirement to the workload. Look for clues about schema, scale, relationships, reporting, real-time processing, and visualization. The correct answer is usually the service family that fits the workload pattern, not the newest product name.